Additional file 3 of Using single-nucleus RNA-sequencing to interrogate transcriptomic profiles of archived human pancreatic islets

Additional file 3: Figure S1. Ambient contribution and islet cell gene marker UMAP and violin plots in scRNA-seq and snRNA-seq in cultured human islets. (A,B) Box plots representing the levels of ambient RNA contamination in droplets of (A) scRNA-seq and (B) snRNA-seq datasets. (C) Violin plot representing the duplication rate in scRNA-seq (orange plot) and snRNA-seq (blue plot) methodologies. The rate was calculated by normalizing the average number of UMI per cell/nucleus on the average number of reads per cell/nucleus. (D, F, H, J) UMAP plots displaying expression levels of (D) GCG, (F) INS, (H) PPY, and (J) SST within the global distribution in scRNA-seq (left panels) and snRNA-seq (right panels). Expression levels are indicated as natural log of counts and range from 0 (gray) to 5 (purple). (E, G, I, K) Violin plots representing expression levels (natural log of counts, Y-axis) of (E) GCG, (G) INS, (I) PPY, and (K) SST within each cluster (X-axis) identified in scRNA-seq (orange plots) or snRNA-seq (blue plots) datasets. Figure S2. Correlation plots and fractional overlap estimation of islet cell types from scRNA-seq dataset with those from the reference dataset. (A) Scatter plots representing correlation of gene expression levels between scRNA-seq-derived (Y-axis) α-cells (first row from top), β-cells (second row from top), PP-cells (third row from top), and δ-cells (fourth row from top) and reference-derived (X-axis) α-cells (first column from left), β-cells (second column from left), PP-cells (third column from left) or δ-cells (fourth column from left). X-axis and Y-axis represent the expression levels in natural log of counts in the indicated datasets. The blue line in each plot represents the regression line, whose fit is indicated by the R2 value (the square of the Pearson correlation coefficient). The red circles indicate the islet cell specific marker genes driving the correlation between the same cell type from the two datasets. The red dotted squares highlight the correlation plots used in the main Fig. 3D. (B-E) Fractional overlap expressed in percentages (%, Y-axis) of the (B) top 100, (C) 200, (D) 500, or (E) 1000 genes between the indicated islet cell types from the reference dataset (X-axis) and the α-cells (yellow bars), β-cells (green bars), PP-cells (blue bars), or δ-cells (red bars) from the scRNA-seq dataset. Figure S3. Quality check parameters in the transplanted human islet snRNA-seq dataset. (A-D) Violin plots of (A) number of genes, (B) number of reads, (C) proportion of mitochondrial genes, and (D) proportion of mouse genes per nucleus across the 4 transplanted human islet samples in the snRNA-seq dataset. (E) Box plot of the levels of ambient RNA contamination in droplets of the in vivo snRNA-seq dataset within each human islet graft. (F) UMAP plot of nuclear cluster distribution and cell type prediction of in vivo snRNA-seq dataset following harmonization to the reference dataset. Figure S4. Islet cell type validation in human islet graft sections by immunofluorescence. Representative images of human islet cells identified as α-cells, β-cells or δ-cells according to the glucagon (GCG, green), insulin (INS, red) and somatostatin (SST, white) labeling. Nuclei are stained in blue. Scale bar is: 50 μm